08-07-2014, 03:04 PM
Correlation-Aware QoS Routing With Differential Coding for Wireless Video Sensor Networks
Correlation-Aware QoS Routing.pdf (Size: 1.86 MB / Downloads: 16)
Abstract
The spatial correlation of visual information retrieved
from distributed camera sensors leads to considerable
data redundancy in wireless video sensor networks, resulting
in significant performance degradation in energy efficiency
and quality-of-service (QoS) satisfaction. In this paper, a correlation-
aware QoS routing algorithm (CAQR) is proposed to
efficiently deliver visual information under QoS constraints by
exploiting the correlation of visual information observed by
different camera sensors. First, a correlation-aware inter-node
differential coding scheme is designed to reduce the amount of
traffic in the network. Then, a correlation-aware load balancing
scheme is proposed to prevent network congestion by splitting the
correlated flows that cannot be reduced to different paths. Finally,
the correlation-aware schemes are integrated into an optimization
QoS routing framework with an objective to minimize energy consumption
subject to delay and reliability constraints. Simulation
results demonstrate that the proposed routing algorithm achieves
efficient delivery of visual information under QoS constraints in
wireless video sensor networks.
INTRODUCTION
RECENT advances in imaging hardware and wireless
communications have fostered the use of video sensors
in various distributed sensing applications [10], [16]. By integrating
imaging sensor, embedded processor, memory, and
wireless transceivers on a single device, a video sensor node is
able to retrieve, process, store, and transmit visual information
under limited power supply. Networks of interconnected video
sensor nodes are referred to as wireless video sensor networks
(WVSNs) [2], [24], in which multiple video sensors collaborate
with each other to provide enriched observations of the
environment. WVSNs can enhance a lot of applications such
as environmental monitoring, traffic enforcement, and remote
health care. Most of these applications require that visual information
be delivered under predefined quality-of-service (QoS)
constraints. This is a challenging task because video sensors
are constrained in battery and processing capabilities, while the
delivery of visual information is resource-demanding.
CORRELATION OF VISUAL INFORMATION
In a densely deployedWVSN, there exists correlation among
the observations from video sensors with overlapped FoV. We
first study the correlation characteristics of visual information in
WVSNs, and then discuss video in-network compression mechanisms
for reducing traffic redundancy
Correlation of Visual Information in Sensor Networks
A video sensor can only observe the objects within its FoV.
As shown in Fig. 1(a), the FoV of a video sensor is determined
by four parameters: the location of the video sensor , the
sensing radius , the sensing direction , and the offset
angle . The sensing process of a video sensor is characterized
by projection from a 3-D scene to a 2-D image, for which
the key parameter is the sensor’s focal length . To simplify
the problem, we consider the case that all the video sensors in
a network are homogeneous (i.e., they share the same values of
, and ). For two sensors and with FoVs and
, suppose at a same time, their observed images are and
, respectively. and are correlated if and are
overlapped with each other. We introduce two metrics to characterize
the correlation between video sensors.
Video In-Network Compression
Due to the huge size of raw visual information, images
and video sequences are compressed prior to transmission. A
lot of standardized techniques can be applied for image and
video coding, such as JPEG/JPEG 2000 and H.26x/MPEG.
These standards are based on the predictive coding concept.
In contrast, distributed video coding (DVC) [11] allows for
separate encoding of correlated sources and joint decoding at
the end user. DVC is introduced to reduce the computational
complexity at the encoders, however, there is a lack of practical
implementations of DVC in sensor networks. On the other
hand, there are many studies on reducing the computational
complexity on low-power DSPs for standardized coding techniques.
For these reasons, we consider the standardized coding
techniques in our work.
Standardized coding techniques can be classified into intra
coding and inter coding. Intra coding reduces the redundancy
within an image, while inter coding (also called differential or
predictive coding) reduces the redundancy among multiple images.
Accordingly, a compressed video sequence usually consists
of periodical intra coded reference (I) frames and inter
coded frames between reference frames. Inter coding has much
higher coding efficiency than intra coding, consequently, intra
coded frames have much larger sizes than inter coded frames.
Intra frames are introduced periodically to reduce the propagation
of packet losses/errors or to start an independent piece of
video stream.
In aWVSN,
CORRELATION-AWARE QOS ROUTING
We propose a correlation-aware QoS routing algorithm
(CAQR) for the delivery of visual information in WVSNs. By
utilizing the correlation characteristics of video sensors, the
algorithm achieves energy-efficient delivery of visual information
while satisfying QoS constraints. CAQR is a distributed
routing solution for WVSNs, and its components are designed
to be implemented on each sensor node. In the following, we
explain the CAQR algorithm in detail
Coding Efficiency in QoS Routing
We now evaluate the gain of correlation-aware coding when
it is implemented in the QoS routing algorithm. We find solutions
to the packet delivery ratio update (PDRU) problem
in Section IV-C, and then we test the best average differential
coding efficiency after channel coding in (35).
The parameters in the PDRU problem (32) are determined as
follows. The average size of an intra frame is determined from
the statistics of the video traces in [23]. The payload length of a
packet is set to 50 Bytes. The number of packets in a frame can
then be estimated from the average size of the frame and the
packet length. We use a series of block codes with structures
[17] for dynamic channel coding. The block length is
set to 127, and the number of correctable bits varies from 1 to
31. A single hop scenario with BPSK modulation is considered,
where the received SNR is assumed to be uniformly distributed
between dB and 15 dB.
The required frame decodable probability is assigned
by specific applications. We set to three different values
here: 0.7, 0.8, and 0.9. The frame decodable threshold DT is
related to the error recovering capability of video decoders.Here
DT is set to 0.75 and 0.9. We let the original differential coding
efficiency vary from 0 to 0.5, and for each combination of
, DT, and , the PDRU problem is solved and the average
differential coding efficiency after channel coding is
obtained.
Correlation-Aware QoS Routing Algorithm
The performance of the proposed routing algorithm is then
evaluated using a distributed network simulator in Java. In a
field of 100 m 100 m, 49 video sensors are deployed in a grid
structure, and a sink node is placed in a corner of the field. The
sensing directions of the video sensors are uniformly chosen so
as to ensure full coverage of the field, and the sensing parameters
of the sensors are given in Table I.
The traffic for the video sensors is generated based on the
features ofWVSN applications.We let a target move in the field
according to the Random Waypoint Mobility model where the
pause time is set to 0. A video sensor is triggered to capture
an image when it detects the target in its FoV. By launching the
target from 10 different locations, we can generate 10 sequences
of events representing different traffic scenarios. The captured
video frames are in QCIF format (176 144), while the size of
CONCLUSION
We have proposed a correlation-aware QoS routing algorithm
for wireless video sensor networks. Based on the correlation
characteristics of visual information in sensor networks, we
introduce a correlation-aware inter-node differential coding
scheme and a correlation-aware load balancing mechanism.
These correlation-aware operations are integrated in a distributed
routing framework. The whole routing algorithm
minimizes energy consumption under delay and reliability
constraints. The performance of the algorithm is evaluated
in terms of energy efficiency, delay performance, and frame
delivery ratio. Evaluation results show that, by integrating correlation-
aware operations in the routing process, the proposed
algorithm achieves efficient delivery of visual information in
wireless video sensor networks.